SERVER-24405 make sharded mapReduce forward collation when input collection is sharded

3 files changed, 89 insertions, 63 deletions

diff --git a/buildscripts/resmokeconfig/suites/sharding_last_stable_mongos.yml b/buildscripts/resmokeconfig/suites/sharding_last_stable_mongos.yml
index 1d75bab8814..7db761001ff 100644
--- a/buildscripts/resmokeconfig/suites/sharding_last_stable_mongos.yml
+++ b/buildscripts/resmokeconfig/suites/sharding_last_stable_mongos.yml
@@ -26,6 +26,8 @@ selector:
     - jstests/sharding/printShardingStatus.js
     # TODO: Change in error reporting, enable when 3.4 becomes 'last-stable'.
     - jstests/sharding/map_reduce_validation.js
+    # TODO Assumes mapReduce can handle the collation option; enable when 3.4 becomes 'last-stable'.
+    - jstests/sharding/mapReduce_inSharded.js
 
 executor:
   js_test:
diff --git a/jstests/sharding/mapReduce_inSharded.js b/jstests/sharding/mapReduce_inSharded.js
index 110be371ba9..52eb379e1fd 100644
--- a/jstests/sharding/mapReduce_inSharded.js
+++ b/jstests/sharding/mapReduce_inSharded.js
@@ -1,62 +1,85 @@
-var verifyOutput = function(out) {
-    printjson(out);
-    assert.eq(out.counts.input, 51200, "input count is wrong");
-    assert.eq(out.counts.emit, 51200, "emit count is wrong");
-    assert.gt(out.counts.reduce, 99, "reduce count is wrong");
-    assert.eq(out.counts.output, 512, "output count is wrong");
-};
-
-var st = new ShardingTest(
-    {shards: 2, verbose: 1, mongos: 1, other: {chunkSize: 1, enableBalancer: true}});
-
-st.adminCommand({enablesharding: "mrShard"});
-st.ensurePrimaryShard('mrShard', 'shard0001');
-st.adminCommand({shardcollection: "mrShard.srcSharded", key: {"_id": 1}});
-
-var db = st.getDB("mrShard");
-
-var bulk = db.srcSharded.initializeUnorderedBulkOp();
-for (j = 0; j < 100; j++) {
-    for (i = 0; i < 512; i++) {
-        bulk.insert({j: j, i: i});
+(function() {
+    "use strict";
+
+    var verifyOutput = function(out) {
+        printjson(out);
+        assert.commandWorked(out);
+        assert.eq(out.counts.input, 51200, "input count is wrong");
+        assert.eq(out.counts.emit, 51200, "emit count is wrong");
+        assert.gt(out.counts.reduce, 99, "reduce count is wrong");
+        assert.eq(out.counts.output, 512, "output count is wrong");
+    };
+
+    var st = new ShardingTest(
+        {shards: 2, verbose: 1, mongos: 1, other: {chunkSize: 1, enableBalancer: true}});
+
+    st.adminCommand({enablesharding: "mrShard"});
+    st.ensurePrimaryShard('mrShard', 'shard0001');
+    st.adminCommand({shardcollection: "mrShard.srcSharded", key: {"_id": 1}});
+
+    var db = st.getDB("mrShard");
+
+    var bulk = db.srcSharded.initializeUnorderedBulkOp();
+    for (var j = 0; j < 100; j++) {
+        for (var i = 0; i < 512; i++) {
+            bulk.insert({j: j, i: i});
+        }
     }
-}
-assert.writeOK(bulk.execute());
-
-function map() {
-    emit(this.i, 1);
-}
-function reduce(key, values) {
-    return Array.sum(values);
-}
-
-// sharded src
-var suffix = "InSharded";
-
-var out = db.srcSharded.mapReduce(map, reduce, "mrBasic" + suffix);
-verifyOutput(out);
-
-out = db.srcSharded.mapReduce(map, reduce, {out: {replace: "mrReplace" + suffix}});
-verifyOutput(out);
-
-out = db.srcSharded.mapReduce(map, reduce, {out: {merge: "mrMerge" + suffix}});
-verifyOutput(out);
-
-out = db.srcSharded.mapReduce(map, reduce, {out: {reduce: "mrReduce" + suffix}});
-verifyOutput(out);
-
-out = db.srcSharded.mapReduce(map, reduce, {out: {inline: 1}});
-verifyOutput(out);
-assert(out.results != 'undefined', "no results for inline");
-
-out = db.srcSharded.mapReduce(
-    map, reduce, {out: {replace: "mrReplace" + suffix, db: "mrShardOtherDB"}});
-verifyOutput(out);
-
-out = db.runCommand({
-    mapReduce: "srcSharded",  // use new name mapReduce rather than mapreduce
-    map: map,
-    reduce: reduce,
-    out: "mrBasic" + "srcSharded",
-});
-verifyOutput(out);
+    assert.writeOK(bulk.execute());
+
+    function map() {
+        emit(this.i, 1);
+    }
+    function reduce(key, values) {
+        return Array.sum(values);
+    }
+
+    // sharded src
+    var suffix = "InSharded";
+
+    var out = db.srcSharded.mapReduce(map, reduce, "mrBasic" + suffix);
+    verifyOutput(out);
+
+    out = db.srcSharded.mapReduce(map, reduce, {out: {replace: "mrReplace" + suffix}});
+    verifyOutput(out);
+
+    out = db.srcSharded.mapReduce(map, reduce, {out: {merge: "mrMerge" + suffix}});
+    verifyOutput(out);
+
+    out = db.srcSharded.mapReduce(map, reduce, {out: {reduce: "mrReduce" + suffix}});
+    verifyOutput(out);
+
+    out = db.srcSharded.mapReduce(map, reduce, {out: {inline: 1}});
+    verifyOutput(out);
+    assert(out.results != 'undefined', "no results for inline");
+
+    // Ensure that mapReduce with a sharded input collection can accept the collation option.
+    out = db.srcSharded.mapReduce(map, reduce, {out: {inline: 1}, collation: {locale: "en_US"}});
+    verifyOutput(out);
+    assert(out.results != 'undefined', "no results for inline with collation");
+
+    out = db.srcSharded.mapReduce(
+        map, reduce, {out: {replace: "mrReplace" + suffix, db: "mrShardOtherDB"}});
+    verifyOutput(out);
+
+    out = db.runCommand({
+        mapReduce: "srcSharded",  // use new name mapReduce rather than mapreduce
+        map: map,
+        reduce: reduce,
+        out: "mrBasic" + "srcSharded",
+    });
+    verifyOutput(out);
+
+    // Ensure that the collation option is propagated to the shards. This uses a case-insensitive
+    // collation, and the query seeding the mapReduce should only match the document if the
+    // collation is passed along to the shards.
+    assert.writeOK(db.srcSharded.remove({}));
+    assert.eq(db.srcSharded.find().itcount(), 0);
+    assert.writeOK(db.srcSharded.insert({i: 0, j: 0, str: "FOO"}));
+    out = db.srcSharded.mapReduce(
+        map,
+        reduce,
+        {out: {inline: 1}, query: {str: "foo"}, collation: {locale: "en_US", strength: 2}});
+    assert.commandWorked(out);
+    assert.eq(out.counts.input, 1);
+})();
diff --git a/src/mongo/s/commands/cluster_map_reduce_cmd.cpp b/src/mongo/s/commands/cluster_map_reduce_cmd.cpp
index b8e7bb9e0c9..b5cf2df1861 100644
--- a/src/mongo/s/commands/cluster_map_reduce_cmd.cpp
+++ b/src/mongo/s/commands/cluster_map_reduce_cmd.cpp
@@ -93,8 +93,9 @@ BSONObj fixForShards(const BSONObj& orig,
 
         if (fn == bypassDocumentValidationCommandOption() || fn == "map" || fn == "mapreduce" ||
             fn == "mapReduce" || fn == "mapparams" || fn == "reduce" || fn == "query" ||
-            fn == "sort" || fn == "scope" || fn == "verbose" || fn == "$queryOptions" ||
-            fn == "readConcern" || fn == QueryRequest::cmdOptionMaxTimeMS) {
+            fn == "sort" || fn == "collation" || fn == "scope" || fn == "verbose" ||
+            fn == "$queryOptions" || fn == "readConcern" ||
+            fn == QueryRequest::cmdOptionMaxTimeMS) {
             b.append(e);
         } else if (fn == "out" || fn == "finalize" || fn == "writeConcern") {
             // We don't want to copy these